@benjaminjkhongSIPA OS: autonomous AI for neurodivergent architects
Hi everyone, I'm Benjamin, a student at Multimedia University in Nongmalaka and a developer. I'm building SIPA OS, which describes itself as autonomous AI for neurodivergent architects. The goal is a clean terminal that cuts cognitive noise, with LLM auditing meant to keep hallucinations out of the output. The code is public on GitHub.
My stack leans on agent and model tooling: the OpenAI Agents SDK, Claude, Mistral, Ollama, Supabase, and Vercel, with Cursor and Claude Code for building. I've also written an Alpaca MCP bot in JavaScript and documented security assessments and HackTheBox labs. SIPA OS is early, and I'm still working out whether the auditing layer holds up in real architecture workflows.
I'd love feedback from anyone who has built evaluation or verification for LLM outputs. How did you decide what counts as a hallucination in your setup, and what did you check first? Thanks for reading.
Comments2
Day 1 begins today :)
The architects part is what I'm curious about. Have you talked with any working architects yet, neurodivergent or not, and what did they say they would actually pay to stop dealing with? That answer might tell you what to check first.
2 comments
Day 1 begins today :)
The architects part is what I'm curious about. Have you talked with any working architects yet, neurodivergent or not, and what did they say they would actually pay to stop dealing with? That answer might tell you what to check first.